Midwest Wheel builds toward AI agents that fix problems
SiliconANGLE Sloane Kali Faye
Midwest Wheel is wiring AI into one system so software can fix routine problems itself. The catch: it’s still making humans check anything tied to cash or bad data.
Based on reporting by SiliconANGLE, Sloane Kali Faye — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Midwest Wheel Companies is taking a fairly blunt approach to AI: keep the company’s capabilities tied to one core system, and make the software fix problems instead of just flagging them. That view came from Steve McEnany, the company’s senior vice president, during a conversation at Infor Velocity Week with theCUBE Research’s Christophe Bertrand and co-host Alison Kosik.
McEnany’s rule is simple enough to fit on a whiteboard. If a new tool gets added, it still has to point back to the Infor system so there is one reference point. He said that matters even when the company brings in a third-party tool. The idea is less about collecting shiny features and more about keeping the plumbing sane.
The move to the cloud helped make that possible. It took server management and software updates off a lean IT team’s plate, and it let the company build needed tools in a matter of weeks. Among the features already in use are a product recommender in order entry and automation that scans emailed PDF invoices into the system.
But McEnany is not handing the keys to the robots. Anything that touches cash or accounts still needs a human in the loop. He also said another AI tool caught a problem in a companywide sales report card that had been built with AI and “didn’t add up.” That is the real story here: AI is useful only if someone checks the math and the rules behind it.
Infor is pushing the same general message, arguing that industry-specific context matters and pointing to research showing that off-the-shelf AI falls short for two out of three businesses. The company says governance has to come first, because bad data and badly written prompts can turn automation into a very expensive typo machine.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of AI story: boring infrastructure, strict guardrails, fewer heroics. The rush to bolt agents onto everything usually ignores the unglamorous part, which is making sure they answer to one system and not a pile of accidental nonsense. In enterprise AI, discipline is the feature the marketing decks keep forgetting to sell.
Read more about this at: SiliconANGLE